IBM releases the LLMON markup language with Granite 4.1 3B and 8B preview models: reserved special tokens separate instructions from data, blocking 47% more injection attacks on OPI than the best frontier model tested
A way of structuring context built for models rather than parsers, as an alternative to JSON, aimed at separating instructions from data at training time. How it works, how much it helps, and where it stops.
// Key points
- LLMON uses six reserved special tokens as structural delimiters, making instructions, data and requirements separately addressable blocks; an exec block states which instruction runs on which data, and instructions not referenced by it are ignored.
- IBM releases two experimental previews, granite-4.1-3b-LLMON-preview and granite-4.1-8b-LLMON-preview, plus a toolkit and lossless JSON-to-LLMON converters, and says general capability is on par with Granite 4.1 Instruct.
- On the OPI indirect prompt injection benchmark it avoids 47% more attacks than the best frontier model tested and 95% more than the best open model, with similar trends on ASB and StruQ.
- IBM says this is learned behavior, not an enforced guarantee, and targets static injections; adaptive attacks that unfold over many turns are still being worked on.
Builder's takeAI Cloud Drive has models read whatever documents users upload, so indirect injection is unavoidable. Small models are already good enough for read-and-extract steps, so I would run the 8B preview on our internal extraction set and measure both injection blocking and extraction accuracy. It is a preview, though; it does not replace least-privilege tools and an outbound allowlist.
// Background · from #Open models
Full timeline →- Nov 13 DeepSeek V4-Pro, V4-Flash and V4.1-Flash arrive on SageMaker JumpStart: MIT-licensed MoE models with a 1M-token context
- Nov 13 SGLang-Neuron is publicly available: serve models and generate RL rollouts on AWS Trainium with the same SGLang interface, including prefill-decode disaggregation
- Nov 12 Microsoft Research discloses Lazarus: reprogramming fewer than 10 dormant neurons, with no training, backdoors an agent built on the model